Intracranial Microvascular Free Flaps
Steven Levine, M.D., Evan S. Garfein, M.D., Howard Weiner, M.D., Michael J. Yaremchuk, M.D., Pierre B. Saadeh, M.D., Geoffrey Gurtner, M.D., Jamie P. Levine, M.D., and Stephen M. Warren, M.D.
Large acquired intracranial defects can result from trauma or surgery. When reoperation is required because of infection or tumor recurrence, management of the intracranial dead space can be challenging. By providing well-vascularized bulky tissue, intracranial microvascular free flaps offer potential solutions to these life-threatening complications. A multi-institutional retrospective chart and radiographic review was performed of all patients who underwent microvascular free-flap surgery for salvage treatment of postoperative intracranial infections between 1998 and 2006. A total of six patients were identified with large intracranial defects and postoperative intracranial infections. Four patients had parenchymal resections for tumor or seizure and two patients had posttraumatic encephalomalacia. All patients underwent operative debridement and intracranial free-flap reconstruction using the latissimus dorsi muscle (n =2), rectus abdominis muscle (n =2), or omentum (n =2). All patients had titanium (n =4) or Medpor (n=2) cranioplasties. We concluded that surgery or trauma can result in significant intracranial dead space. Treatment of postoperative intracranial infection can be challenging. Vascularized free tissue transfer not only fills the void, but also provides a delivery system for immune cells, antibodies, and systemically administered antibiotics. The early use of this technique when intracranial dead space and infection coexist is beneficial.
KEYWORDS: Microvascular free flap, intracranial, infection, reconstruction, salvage
The intracranial compartment is normally occupied, almost in its entirety, by the brain and the meninges. Following surgical or incidental trauma, the brain volume may be diminished acutely (e.g.,tumor resection) or subacutely (e.g., encephalomalacia following trauma). The space between the remaining brain and the cranium is eventually filled with fibrinous material and fluid. Unfortunately, this proteinaceous fluid provides a fertile substrate for bacterial growth, especially when the frontal sinus is comminuted, the tissues are irradiated, or there is a cutaneous deficit.
Rebuilding the skull after trauma or parenchymal resection can be challenging, particularly when the autogenous or alloplastic construct spans a large intracranial dead space. This is especially true following infection. Surgeons have attempted to solve this problem in a variety of ways. For example, White et al described a reduction in the extradural dead space by mobilizing and reattaching the dura to the vault floor. Rasmussen et al performed a hemispherectomy but left devitalized portions of brain tissue as an intracranial filler. Unfortunately, both techniques predisposed the patient to infection. Ultimately, surgeons have developed guiding principles for intracranial reconstruction: obliteration of intracranial dead space, covering vital structures with well-vascularized tissue, sealing dural tears, and cranializing the frontal sinus.
In 1993, Matheson et al combined an Adams-type subdural space reduction with an extradural microvascularly transferred omental flap. Sugawara performed a similar extradural omental flap. More recently, Schwabegger et al performed a two-stage myocutaneous microvascular reconstruction of a massive hemispheric defect. Here, we review our experiences with microvascular free-flap (MVFF) salvage of intracranial infections in the setting of large dead space lesions. Three clinical cases are provided to highlight different types of MVFFs that are designed not only to reduce intracranial dead space, but also facilitate alloplastic calvarial reconstruction.
A retrospective chart review was performed to identify all patients who underwent intracranial MVFF reconstruction at the New York University Medical Center and the Massachusetts General Hospital from 1998 to 2006. Four patients had parenchymal resection (n=1) subtotal and n=3 total hemispherectomy) for tumor (n=2) and seizure (n=2) treatment, and two had sustained head trauma in motorcycle accidents. All six patients developed postoperative intracranial infections resulting in calvarial bone graft loss and massive surface contour irregularity.The choice of donor site (latissimus dorsi versus rectus abdominis versus omentum) was based on past medical history, preoperative clinical examination, radiographic findings, and severity of soft tissue deformity. Frustra or cone-shaped defects, often deep set with a limited intracranial access point and little or no scalp tissue deficiency, were preferentially reconstructed with rectus MVFFs. Hyperboloid, cylindrical, or spherical intracranial defects that required broad cutaneous coverage were reconstructed with myocutaneous latissimus dorsi MVFFs. Defects that were multiforme with complex three-dimensional (3D) architecture, which involved the frontal sinus and did not require cutaneous coverage, were reconstructed with the omentum. Persons with a past medical history or clinical examination that precluded first-choice flap harvest (e.g., Case 2) were treated with an alternate flap.
In brief, each free flap was harvested in standard fashion. We used the deep inferior epigastric artery and vein for the rectus abdominis muscle flap (n=2) and the thoracodorsal artery and vena comitantes for the latissimus dorsi muscle flap (n=2). The right gastroepiploic artery was used as the donor vessel in both omental flaps (n=2). Recipient vessels included the superficial temporal artery (n=5) and vein (n=4), the facial artery (n=1) and vein (n=1), and the external jugular vein (n=1). The choice of recipient vessels was made based on relative size, ease of dissection, and proximity to the cranial defect.
Because the three MVFF donor tissues had relatively large-caliber donor vessels, the superficial temporal artery and vein were the primary recipient vessels. When the superficial temporal artery was hypoplastic (Case 2), the facial artery was chosen as an alternate arterial supply (n=1). When the superficial temporal vein was thrombosed (n=1) or small (n=1), the facial vein (n=1) and the external jugular vein (n=1) were selected as alternate venous drainage systems. One patient required a 10-cm saphenous vein graft from the inferior epigastric donor vessel to the external jugular vein because the superficial temporal vein was thrombosed. Donor artery and veins with or without vein grafts were always tunneled subcutaneously.
QB was a 56-year-old man who underwent combined transcranial and transnasal excision of a subfrontal anaplastic meningioma in 1998. In 2004, he underwent a second transcranial resection for recurrent tumor. During this second operation, tumor invading the skull base was resected, and the fovea ethmoidalis and cribriform were reconstructed with a fat graft, BioGlue (CryoLife Inc., Kennesaw, GA), and a vascularized pericranial flap. The postoperative course was complicated by Acinetobacter baumannii meningitis. On postoperative day 20, the patient returned to the operating room for drainage of a left anterior cranial fossa intracranial abscess and debridement of devitalized tissue. Postoperatively, the patient had a persistent cerebrospinal fluid (CSF) leak, anterior cranial defect, and a scalp contour deformity. A rectus abdominis free muscle flap was designed to address the persistent CSF leak, contour deformity, and refractory infection (Fig. 2). The rectus MVFF was inset into the frustra-shaped defect, and the inferior epigastric artery and vein were anastomosed to the superficial temporal artery and facial vein, respectively. The patient did well postoperatively; the CSF leak stopped, the intracranial infection resolved, and the scalp contour deformity was corrected with a mesh cranioplasty. Twenty-two months later the patient remained disease-free, without CSF leak or intracranial infection despite a small persistent dead space.
JH had an autoimmune disorder (Rasmussen’s syndrome) characterized by progressive neurological deterioration and seizures. The patient underwent a right hemispherectomy for intractable seizures and placement of a ventriculoperitoneal shunt at 8 years of age. One year later, he developed an infection that necessitated removal of the calvarial bone graft. He subsequently underwent two alloplastic reconstructions; both implants became infected and were removed. At age 12, he presented with chronic osteomyelitis and an open 12 x 12-cm right frontoparietal calvarial contour deformity. Reconstruction of this broad surface intracranial defect was planned with a latissimus dorsi myocutaneous free flap. The latissimus dorsi myocutaneous flap was selected to fill the intracranial defect and provide vascularized coverage for the alloplastic cranial reconstruction. A latissimus dorsi myocutaneous flap was harvested and the muscle inset into the defect. The pedicle was tunneled subcutaneously and anastomosed to the ipsilateral facial artery and vein. The skin paddle was used to reconstruct the scalp defect. On postoperative day 4, the flap was congested. Returning to the operating room, a small (<25 mL) hematoma was evacuated from the anastomotic site. A small portion(~16 cm2)of the skin paddle was excised and closed primarily. The patient was discharged to home on post-operative day 11. One year after latissimus dorsi MVFF reconstruction, the patient remained infection-free with stable scalp reconstruction. Case 3
AP was a 42-year-old man who was involved in a high-speed motorcycle crash in 2005. At the time of his injury, he underwent emergent right cranial decompression via a frontoparietal craniotomy. The bone flap became grossly contaminated during ex vivo storage and was discarded. The patient presented to the plastic surgery clinic with a large right-sided intracranial defect, a calvarial contour irregularity, intracranial communication with the frontal sinus, and malunion of the right frontal bone and zygoma. His right frontal bone, right orbital roof, superior orbital rim, and zygoma were osteotomized, anatomically reduced, and fixed with 1.0-mm titanium plates (Synthes, West Chester, PA). The right frontal sinus was sealed with a galea frontalis flap. Alloplastic cranioplasty was planned as a second stage.
In the postoperative period, the patient had a persistent CSF leak and developed Escherichia coli meningitis. The devitalized galea frontalis flap and frontal bone graft were debrided. After his condition stabilized, a free omental flap was placed in the defect, covering not only dura but also filling the frontal sinus. The right gastroepiploic artery and vein were anastomosed to the right superficial temporal artery and vein. A vacuum-assisted closure dressing was employed over the flap for several days until the skin flap edema subsided. Eventually, a delayed primary closure was performed. The patient had an uneventful postoperative course. A custom-designed Medpor implant (Proex Inc., Newman, GA) was placed to reconstruct the cranial defect 6 months later.
Cranioplasties can provide excellent aesthetic and functional results following surgical or incidental trauma. Cranioplasties that span large intracranial defects are prone to failure due to the poor vascularity of the surrounding tissues and the fluid-filled cavity. When the cranial bone graft is lost, differential intracranial and external atmospheric pressure can result in contour deformities, skin ulceration, and cerebral hypoperfusion. The study presented here demonstrates the advantages of free tissue transfer as salvage treatment for large cranial defects, which include (1) obliteration of the dead space, (2) a conduit for immune cells, antibodies, and systemically administered antibiotics, and (3) assistance in restoring volume and contour. More- over, the availability of safe and reliable reconstructive options may improve the confidence of the extirpative neurosurgeon and the soft tissue and bony debridement by the assisting plastic surgeon.
Several MVFF donor options are available for intracranial reconstruction. In the present study, we described the use of the latissimus dorsi, rectus abdominus, and omentum. Each of these flaps have advantages that make them suitable for intracranial reconstruction following infection. Muscle or myocutaneous flaps offer an excellent source of vascularity, and their salvage efficacy in lower extremity infection is well known. The flaps described in this report are also highly reliable and relatively easy and quick to harvest. Interestingly, 3D analysis of images obtained from patient QB 22 months postoperatively demonstrate a persistent dead space, indicating the critical element of the salvage procedure may be the delivery of well-vascularized tissue to zone of infection rather than the absolute elimination of dead space. The use of omental free flaps to fill extradural dead space intracranially and in the spine has also been described. In addition to its immunologic and neoangiogenic properties, the omentum has a dense lymphatic network with tremendous absorptive potential.
The timing of cranioplasty is variable and depends on the medical condition of the patient and control of the processes that led to free-flap reconstruction of the intracranial dead space. In principle, alloplastic material should not be introduced into a contaminated field; instead, the implant should be reintroduced after a 3- to 6-month infection-free period. However, four of six patients in this series had titanium mesh placed to reconstruct their cranial defects at the same time as the MVFF reconstruction; none of these patients suffered post-MVFF infections. The decision to introduce titanium mesh into an infected field was difficult but successful in this series.
Cranioplasty can be accomplished using autologous bone or alloplastic implants. Titanium mesh is relatively inexpensive. Medpor implants can be customized to patient-specific needs based on preoperative computed tomography data. In this series, four patients were treated with titanium mesh and two patients received Medpor implants. The alloplastic reconstructions subjectively improved the head shape in all six patients. In sum, the risk of alloplastic infection appears to be low when the intracranial dead space is reduced and new vascularized tissue is delivered to the wound.
The choice of donor site (latissimus dorsi versus rectus abdominis versus omentum) flap is based on past medical history, preoperative clinical examination,availability of donor tissues, radiographic findings, defect geometry, and severity of the soft tissue deformity. For example, patients with a history of multiple prior abdominal operations, intraperitoneal infections (e.g., diverticulitis, cholecystitis, pancreatitis), omentectomy, young children (e.g., Case 2), or cachexia would not be good candidates for an omental flap. Moreover, many patients with significant intracranial injury have had gastrostomy tubes placed. Although these are not absolute contraindications to an omental flap, they make harvest more difficult. In patients with hemiplegia, a muscle flap from the ipsilateral affected side should be considered (e.g., Case 2). Frustra or cone-shaped defects, often deep set with a limited intracranial access point and little or no scalp tissue deficiency, can be reconstructed with a rectus MVFFs. Hyperboloid, cylindrical, or spherical intracranial defects that required broad cutaneous coverage can be reconstructed with a broad myocutaneous flap such as latissimus dorsi. Defects that are multiforme with complex 3D architecture, which involve the frontal sinus and do not require cutaneous coverage, are suitable for omental reconstruction.
Recipient vessels choice was based on availability and suitability. All flaps described in this report had relatively large-caliber vascular pedicles. Therefore, the superficial temporal artery (n=5) and vein (n=4) were chosen as primary recipient vessels. When the superficial temporal artery was unavailable, the facial artery was selected as an alternate arterial supply (n=1). Alternative choices for the superficial temporal vein include the facial vein (n=1) and the external jugular vein (n=1). Only one patient required a vein graft to bypass the superficial temporal vein for the external jugular vein. Donor artery and veins with or without vein grafts were always tunneled subcutaneously.
Free tissue transfer should be considered as part of the armamentarium for reconstruction and salvage of large infected intracranial wounds. The MVFF fills the intracranial deep space and delivers new blood to the infected area, making alloplastic reconstruction of the calvarium possible.
We hereby certify, that to the best of our knowledge, no financial support or benefits have been received by any coauthor, by any member of our immediate family or any individual or entity with whom or with which we have a significant relationship from any commercial source that is related directly or indirectly to the scientific work reported on in the article.